1. Field of the Invention
The present invention relates to image processing apparatuses and methods, and recording media and programs used therewith. In particular, the present invention relates to an image processing apparatus and method that have increased accuracy of motion vector detection, and to a recording medium and program used therewith.
2. Description of the Related Art
In an image processing apparatus for converting the frame frequency of an image, the gradient method is used as a method for detecting a motion vector of a block of interest. Although the gradient method can produce a highly accurate result of detection for a minute movement, it is not practical because, when a movement in an actual moving image is detected, such a movement is too large. Accordingly, as disclosed in Japanese Unexamined Patent Application Publication No. 60-158786, detection is achieved by recursively performing gradient operation.
However, simply recursively performing the gradient operation takes a long time in the case of performing real-time processing. Accordingly, when performing the gradient operation, from among peripheral block motion vectors detected in past frames, etc., an initial vector having a similar movement is found. By using the initial vector as an initial value for the gradient method, the number of times the gradient operation is performed is reduced. The initial vector can be found, based on spatial or temporal correlation of moving objects, from motion vectors such as a motion vector in a peripheral block around a block of interest, and a motion vector in a block having a phase identical to that of a block of interest in a previous frame.
However, in an object having a large amount of movement, or in a block in the vicinity of a boundary between different moving objects, the optimal motion vector (i.e., a motion vector similar to calculated movement) does not exist which is used as the initial vector in the gradient method. Accordingly, a problem occurs in that a false initial vector (initial vector having different movement) is selected. In addition, this results in a decrease in the accuracy of motion vector detection in the subsequent stage by the gradient method, thus causing image deterioration.